their possible adverse impact on people and the environ-ment. As a consequence, the “Committee for health-related evaluation of building products“ (Ausschuss zur gesundheit-lichen Bewertung von Bauprodukten – AgBB) has developed an evaluation system for testing the emission behaviour of construction products. In this system, the chamber air con-centrations of volatile and semi-volatile organic compounds (VOCs and SVOCs) are determined after 3 and 28 days with the standard thermal desorption method using Tenax ad-sorption tubes [1]. Very volatile organic compounds (VVOC) with boiling points in the range of < 0 to 100 ° C are not deter-mined quantitatively, because Tenax is not appropriate for VVOCs. The VVOC class includes substances such as metha-nol (MeOH), ethametha-nol (EtOH), acetone, 2-chloropropane and pentane. The two primary alcohols methanol and ethanol were chosen as VVOC examples for the method develop-ment. Some of the properties of methanol and ethanol are shown in Table 1.
Methanol and ethanol are high production volume che-micals. They are, for example, used as starting materials for the synthesis of other basic chemicals and also as solvents for resins, paints and oils [3].
A parquet adhesive based on silane-modified polymers, a so-called SMP adhesive, was selected for the practical testing of the method. According to the manufacturer, these adhesives are water- and solvent-free and consist of only one compo-nent. While absorbing water, the polymer chains, such as polypropylene glycol with methoxy-silane groups in a telechelic (terminal) configuration, condense into a three- dimensional network and cleave off methanol [4].
2 Problem
The established measuring methods for the determination of VOC and SVOC emissions based on thermal desorption of Tenax sorbent tubes [5] are not suitable for VVOCs. The Ger-man Research Foundation [6] has published two methods for quantifying methanol in workplace air. These are based on the adsorption of methanol on water (method 1), or on the adsorption on silica gel and extraction in water (method 2), followed by separation with gas chromatography (GC) and detection using a flame ionization detector (FID). Both methods have the disadvantage that they require high sam-ple volumes of 10 l (method 1) and up to 20 l (method 2). In addition, the determination limit of 4.0 mg/m³ for method 1 is very high. For method 2 it is significantly lower at
Development of a TDS-GC-FID method for the
determination of methanol and ethanol in air
Abstract A method for the determination of very volatile organic com-pounds (VVOC) in air was developed and successfully tested for metha-nol and ethametha-nol. Carbotrap 300® (a multi-bed tube using Carbopack C®, Carbopack B® and Carbosive SIII®) was selected as sorption mate-rial for thermal desorption and tested for its suitability for emission test chamber measurements. The adsorption behaviour and the chromato-graphic results are influenced negatively by spiking aqueous solutions on the adsorption medium for calibration purposes. The latter effect can be reduced by blowing dry air across the adsorbent to remove the adsorbed water. Detection and quantification limits were determined for a GC-FID system and compared to GC-MS. The applicability of the method was tested with a methanol eliminating silane modified polymeric adhesive (SMP-adhesive for parquet) in an emission test chamber.
Entwicklung einer TDS-GC-FID-Methode zur
Bestimmung von Methanol und Ethanol in Luft
Zusammenfassung Eine Methode zur Bestimmung von sehr leicht flüch-tigen Verbindungen in Luft wurde für Methanol und Ethanol entwickelt und erfolgreich getestet. Dafür wurde Carbotrap 300® (Multibettrohr aus Carbopack C®, Carbopack B® und Carbosive SIII®) als Sorptions-medium für die Thermodesorption ausgewählt und dessenEignung für Emissionskammermessungen untersucht. Das Adsorptions-verhalten und die chromatographischen Ergebnisse werden negativ durch die Aufgabe von wässrigen Kalibrierlösungen auf das Adsorptions-medium beeinflusst. Durch das Überleiten von trockener Luft kann dieser Effekt verringert werden, da so das adsorbierte Wasser entfernt wird. Die Nachweis- und Bestimmungsgrenzen wurden für ein eingesetztes GC- FID-System ermittelt und mit einem GC-MS verglichen. Mit einem methanolabspaltenden silanmodifizierten Polymerklebstoff (SMPPar -kettkleber), der in eine Emissionskammer eingebracht wurde, konnte die Tauglichkeit der Methode überprüft werden.
A. Pech, O. Wilke, B. Mull, W. Horn, O. Jann
Dipl.-Lebensmittelchem. Anja Pech (Food chemist), Dr. rer. nat. Olaf Wilke, Dr. rer. nat. Birte Mull, Dr. rer. nat. Wolfgang Horn, Dr.-Ing. Oliver Jann, BAM Federal Institute for Materials Research and Testing, Berlin, Germany.
1 Introduction
With increasing environmental and health awareness, the demands on building materials also increase. They should be made from low-emission components so as to minimize
Molar mass in g/mol Melting point in °C Boiling point in °C Density in g/cm³ Vapour pressure 20 °C in mbar log Kow Methanol 32.04 -98 65 0.79 129 -0.74 Ethanol 46.07 -114 78 0.79 58 -0.3
0.065 mg/m³, but requires a much longer sampling time of about 8 h. These two methods are therefore only suitable for emission analysis under these conditions.
The goal is to develop an active sampling method with rela-tively small sample volumes of 1 to 5 l and significantly lower limits of quantification. The application of thermal desorp -tion is most promising, since no solvents are needed and there fore no dilution takes place [7]. Different adsorbents were tested on their ability to sorb methanol and ethanol [8]. As potentially suitable sorbents, Carbosieve SIII®, Carboxen 1000®, Carboxen 1003® and Carboxen 569® were chosen. The 3-bed combination tube Carbotrap 300® is a combina -tion of Carbopack C®, Carbopack B® and Carbosieve SIII®. The suitability of the selected sorbent was tested in emission test chambers. As the method according to DIN ISO 16000-6 specifies sampling of humid air ((50 ± 5) % relative humi -dity), the influence of air humidity was also examined.
3 Materials and methods
For the preparation of standard solutions, the following che-micals were used: methanol (98.8%) from J. T. Baker (for or-ganic residue analysis) and ethanol (99.8%) from Appli-chem. The solvent was water, purified with Easy Pure II® sys-tem from Werner. The resistivity of the purified water was 17.9 MWcm.
For identification and quantification of the substances in the emission test chamber air, a TDS-GC-FID system was used and its performance compared to a TDS-GC-MS system (see Section 3.2). The FID measurement system used was a gas chromatograph, type 6890, Agilent, with a Zebron WAX® Plus (polyethylene glycol) column (30 m × 0.32 mm × 0.5 µm). Helium was used as the carrier gas. The coupled thermal desorption unit was a TDSA with the cold injection system CIS 4, Gerstel, which was cooled to - 100 °C with liquid nitro-gen. A type 6890 gas chromatograph from Agilent with a Res-tek Rxi 5 separation column (60 m × 0.25 mm × 0.25 µm) was used as the second analysis system. This was coupled with an Agilent 5973 N mass selective detector and a thermal desorp-tion system (TDSA 2 with cold injecdesorp-tion system) from Gers-tel. The following temperature programme was used for both thermal desorption units: 40 °C, at 12 °C/min to 200 °C for 0.40 min. The temperature pro gramme of the cold injecti-on system goes through the follow ing steps: - 100 °C for 0.70 min, at 12 °C/min to 200 °C for 3 min. After a non-split injection onto the corresponding separation column, the fol-lowing column temperature programme was run: 40 °C, at 20 °C/min to 220 °C for 5 min.
In scan mode, masstocharge ratios of 25 to 200 were con -sidered. In the selected ion monitoring (SIM) mode, the three
most intense masses 29, 31 and 32 were selected for metha-nol, and for ethanol the masses 29, 31, 45 and 46.
3.1 Sorbent comparison
To compare the sorbents, three tubes with Carbosieve SIII®, Carboxen 1000®, Carboxen 1003®, Carboxen 569® and four tubes with Carbotrap 300® were spiked with 1 µl of a stan-dard (c = 700 ng/µl). The water used as solvent was removed with 1 l of dry synthetic air and the tubes were then analysed with the TDS-GC-FID. The Carbotrap 300® tubes were com-mercial products. The other tubes were prepared by filling empty tubes (Gerstel) manually with 300 mg Carbosieve SIII®, Carboxen 1000®, Carboxen 1003® and Carboxen 569®. The main characteristics of the different tubes are summa -riz ed in Table 2. All sorbents are marked as hydrophobic by the manufacturer.
3.2 Characteristics of the methods
Three different concentrations (20, 200 and 2,000 ng/µl) were injected into the Carbotrap 300® tubes and then ana -lyzed with the TDS-GC-FID and TDS-GC-MS systems, using the same TDS programme. The limits of detection and quan-tification for the FID system were determined according to DIN 32645 [10] with calibration points in the lower mass range from 1 to 5 ng, whereas they were estimated for the MS system using the three point calibration.
The effect of water on adsorption capacity was examined in quadruplicate. For this, the tubes were rinsed with 1 l of dry air or 1 l of humid air ((50 ± 5) %) after manual standard in-jection of 200 ng of methanol and ethanol in water.
3.3 Recovery experiments in the 1 m3 emission test chamber To examine the applicability of the measuring method in emission test chambers, recovery experiments were per -formed. For this purpose two variants were tested. The variant 1 represents the sampling method according to DIN EN ISO 16000-9 (without removing water). In variant 2, 1 l of dry air was passed over the tubes after sampling, re -moving the water bound to the sorbents. For both variants 1 µl of a standard aqueous solution (c = 80 µg/µl) was in-jected into the emission test chamber and then a 1 l sample was taken. Assuming that methanol and ethanol evaporate completely into the gas phase, the concentration in the emis-sion test chamber should be 80 µg/m3. A temperature of (23 ± 1) °C and a relative humidity of (50 ± 5) % were selected as chamber parameters according to DIN EN ISO 16000-9. The air change rate was 0.4/h. The associated calibration functions for methanol and ethanol, for which the Carbotrap 300® tubes were also treated according to variants 1 and 2, are shown in Figures 1 and 2.
Carbosieve SIII® Carboxen 1000® Carboxen 1003® Carboxen 569®
Bulk density in g/cm³ 0.66 0.47 0.49 0.61
Granular size in mesh 60 to 80 60 to 80 40 to 60 20 to 45
Pore diameter in Å 4 to 11 10 to 12 5 to 8 5 to 8
Macro porosity in cm/g 0 0.25 0.28 0.1
Meso porosity in cm/g 0.04 0.16 0.26 0.14
Micro porosity in cm/g 0.35 0.44 0.38 0.2
Surface area in m²/g 975 1 200 1 000 485
3.4 Emissions from an SMP adhesive
The emissions from the SMP adhesive were measured in two experiments. In the first expe-riment, approximately 120 g were applied on each of three glass plates sized 54 cm × 25 cm, with an individual area of 0.1196 m² (minus 1 cm for the edges) and a total area of 0.3588 m² (three plates). Taking into account the ex-change of air, the area-specific air flow rate is q = 1.67 m³/m²h. Using a doctor blade, the material was spread on the glass plates, keeping a 1 cm distance from the edges. Then the glass plates were placed in a 1 m3 chamber of type VCE 1000 classic, Vötsch Industrietechnik. A temperature of (23 ± 1) °C and a relative humi-dity of (50 ± 5) % were selected as chamber parameters accord ing to DIN EN ISO 16000-9. The air change rate was 0.6/h. The emission test chamber air was sampled three times a day over the next nine days.
The second experiment was a variation in the size of the glass plate. For this purpose, a glass plate sized 27 cm × 22 cm with a total area of 0.05 m2 (with a 1 cm margin) and about 50 g of glue was used. Taking into account the ex-change of air, the area-specific air flow rate was q = 12 m³/m²h. The emission test chamber air was sampled three times daily for seven days. The summary of experimental parameters is shown in Table 3.
4 Results
4.1 Sorbent comparison
Figure 3 shows the adsorbed amounts of methanol and ethanol (relative values) for the different sorbents. The standard deviations of the triplicate or quadruplicate determinations are shown as error bars. The comparison of the sorbents shows that all tubes used exhibit a sig-nificantly higher affinity to ethanol than to methanol. The highest adsorption capacity with respect to the target substance methanol was ex hibited by Carbotrap 300® and Carboxen 1000®. Taking into account the corresponding
standard deviations, there is no significant difference be -tween these two sorbents. Carboxen 569® however, has the lowest absorption capacity with respect to methanol and ethanol. Furthermore, the adsorption capacity of the sor-bents with respect to ethanol should also be noted. In that regard Carbotrap 300® adsorbs more ethanol than Carboxen 1000® – therefore the combined sorbents of Carbotrap 300® became the material of choice.
4.2 Characteristic values of the methods
Comparison of the scan and SIM measurements on the MS system shows that, as expected, the SIM method is more sen-sitive than the scan method, even if only slightly (Figure 4). Based on the three concentrations (20, 200 and 2,000 ng/µl) the limits of detection and quantification in the SIM were es-timated taking the smallest mass of 20 ng as LOD (Table 4). The corresponding values for a measurement uncertainty of Figure 1. Calibrations of methanol according to variant 1 (wet) and 2 (dry).
Figure 2. Calibrations of ethanol according to variant 1 (wet) und 2 (dry).
Glass plate Adhesive Emission chamber
Experiment Number of glass plates
Size Mass per
plate in g Area per plate in m² Total area in m² Air change rate in 1/h Loading in m²/m³ Area specific air change rate in m³/m²h 1 3 54 cm × 25 cm 120 0.12 0.36 0.6 0.36 1.67 2 1 27 cm × 22 cm 50 0.05 0.05 0.6 0.05 12
33.3% and level of error of 5% for the two substances metha-nol and ethametha-nol on the FID system are summarized in Table 5. For both methanol and ethanol, the limits of detection and quantification obtained were much lower than the safety
limits required by the German Research Foundation (MAK values). Comparison of de-tection and quantification limits of the FID with those of the MS (in SIM mode) shows that the FID system is much more sensitive (see Tables 4 and 5). The comparison also shows that the column used in the standard VOC analysis composed of 5% phenyl- and 95% methylpolysiloxane features much poorer chromatographic separation.
The impact of humidity on sensitivity is shown in Figure 5. For methanol and ethanol, a marked reduction of the peak areas is evident after blowing through 1 l of humid air com -pared to 1 l of dry air. However, for methanol the effect is greater by a factor of 3 and for ethanol by a factor of 2. The ab solute standard devia tion is similar for both variants, and the-refore dependent on the measuring system.
4.3 Recovery experiments in the 1 m3 emission
test chamber
The results for variants 1 and 2 are shown in Figure 6. Both variants provide good recovery rates for methanol and ethanol. In both expe-riment set-ups, the recovery rates for methanol are higher than the corresponding re -covery rates for ethanol – 110% for variant 1 and 115% for variant 2 for methanol, and 92% for variant 1 and 102% for variant 2 for etha-nol. In general, recovery rates for variant 2 are higher than those of variant 1. In addition, standard deviation of the results of variant 2 for methanol is not as high as in variant 1. The fact that higher recovery rates can be achieved for variant 2 is surprising because a total of 2 l (1 l of humid air and 1 l of dry air) were passed over the tube in this variant. It was expected that the additional volume of air would cause small losses of methanol and ethanol. However, additional results from experiments carried out in parallel [11] con -firmed the expectation – leaving the signifi-cantly improved chromatographic separation as the main cause. By flushing the tubes with an additional litre of dry air, the water which had been adsorbed could be removed. This results in much narrower signal peaks (higher resolution) for methanol. This experiment also confirms the assump -tions in Section 4.2. Because of the gaseous injection of the methanol, much less water sorbs on the sorbents. The inter-ference from water is therefore less pronounced. Therefore, the gaseous injection (e.g. by a gas-collecting tube) is prefe-rable over liquid injection [11]. One option for gas phase calibration is provided by a gas mixing system as developed by Richter et al. [12].
4.4 SMP adhesive
The two test runs differed in the corresponding load q of the emission test chamber (Table 3). The comparison of the ex-perimental results was made on the area-specific emission rates SERA. The methanol emissions measured in the first run were well above the calibration range in the first five days. In order to better investigate the emission behaviour in Figure 3. Sorbent comparison.
Figure 4. Comparison of scan and SIM for methanol and ethanol.
Compound LOD in µg/m3 LOQ in µg/m3
MeOH 20 60
EtOH 20 60
Table 4. Estimated limits of detection (LOD) and limits of quantification (LOQ) for the MS system.
Compound LOD in µg/m3 LOQ in µg/m3
MeOH 1 l dry air 2 8
EtOH 2 9
MeOH 1 l wet air 3 15
EtOH 3 15
Table 5. Limits of detection (LOD) and limits of quantification (LOQ) for the FID system.
the initial phase, another test run with a lower load followed, during which methanol quanti-fication could be performed after only one day. The time-based progression of the area-specific emission rates is shown in Figure 7. Because the calibration range was exceeded, there is no data available for the higher load -ing (large plate) at the start of the measure-ment. A significant decrease of methanol emissions is apparent over the test period. The decrease features a drop which is typical for highly volatile compounds which have no sink effects. This also makes plain a major problem of VVOC analysis with respect to emissions. Due to the high vapour pressure, a rapid transition to the gas phase takes place, accompanied by high initial SERAs which rapidly decreases in this case. To capture this pronounced dynamic, the analysis must be performed over a short period of time. This can be accomplished with the method developed.
5 Conclusion
The experiments show that methanol and ethanol can be quantified by thermal desorp-tion using sorbents already available on the market. The sensitivity of a flame ionization detector is much higher than that of a single quadrupole instrument. However, the in -fluence of humidity is problematic. Despite the hydrophobic properties of the sorbents, water bonded and impaired the chromato -graphic separation – which was alleviated by subsequently passing dry air through the sor-bent. At 15 µg/m³ the estimated limit of quan-tification for methanol is much lower than using previously developed methods, despite poorer chromatographic separation. In addi -tion, the thermal desorption method requires much smaller sample volumes and a reduced workload. The investigations of the SMP ad-hesive in a 1 m3 emission test chamber gave first results of methanol emissions from these adhesives. Emission rates between 250 and 1,200 µg/m²h for methanol were determined after seven days.
Figure 5. Comparison of blowing off by dry and wet air.
Figure 6. Recovery rates in the 1 m³ chamber.
Figure 7. SERA of methanol for the SMP adhesive as a function of time.
Acknowledgments
We thank the Federal Environmental Agency (UBA) for their financial support under grant number 371095305. Our warm thanks go to Sandra Walther for her contribution.
References
[1] Vorgehensweise bei der gesundheitlichen Bewertung der Emis-sionen von flüchtigen organischen Verbindungen (VOC und SVOC) aus Bauprodukten. Ed.: Ausschuss zur gesundheitli-chen Bewertung von Bauprodukten (Committee for Health-re-lated Evaluation of Building Products: Health-reHealth-re-lated Evaluati-on of EmissiEvaluati-ons of Volatile Organic Compounds (VOC and SVOC) from Building Products). Berlin 2010.
[2] GESTIS-database on hazardous substances. Ed.: Deutsche Gesetzliche Unfallversicherung, Berlin. www.dguv.de/ifa, Webcode e20602
[3] Römpp Chemie Lexikon. Stuttgart: Georg Thieme 2012. www.roempp.com/prod/
[4] Mikaelsson, H.; Motzet, H.: Elastische Parkettklebstoffe – Die Erfolgsgeschichte einer Innovation. Ed.: Gesellschaft Deutscher Chemiker, Frankfurt am Main, 2011.
www.aktuelle-wochenschau.de/2011/w39/woche39.html [5] DIN ISO 16000-6: Indoor air pollution – Part 6: Deter mination
of volatile organic compounds in indoor and test chamber air by active sampling on Tenax TA® sorbent, thermal desorption and gas chromatography using MS or MS-FID. Berlin: Beuth 2012.
[6] Analytische Methoden zur Prüfung gesundheitsschädlicher Arbeitsstoffe (Analytical Methods for the Investigation of the Health Hazards of Chemical Compounds). Vol. 1: Air analyses. Ed.: Deutsche Forschungsgemeinschaft (German Research Foundation). Weinheim: Wiley-VCH 1984.
[7] Woolfenden, E.: Sorbent-based sampling methods for volatile and semi-volatile organic compounds in air – Part I
Sorbent-based air monitoring options. J. Chromatogr. A 1217 (2010), p. 2674-2684.
[8] DIN EN ISO 16017-1: Innenraumluft, Außenluft und Luft am Arbeitsplatz – Probenahme und Analyse flüchtiger organischer Verbindungen durch Sorptionsröhrchen/thermische Desorp -tion/Kapillar-Gaschromatographie-Teil 1: Probenahme mit einer Pumpe (Indoor, ambient and workplace air – Sampling and analysis of volatile organic compounds by sorbent tube/ thermal desorption/capillary gas chromatography – Part 1: Pumped sampling). Berlin: Beuth 2001.
[9] Carbon Physical Characteristics. Ed.: SIGMA-Aldrich. www. sigmaaldrich.com/content/dam/sigmaaldrich/docs/ - Supelco/General_Information/1/carbon_physical_characteris-tics.pdf
[10] DIN 32645: Chemische Analytik – Nachweis-, Erfassungs- und Bestimmungsgrenze unter Wiederholbedingungen – Begriffe, Verfahren, Auswertung (Chemical analysis – Detection limit, decision limit and determination limit under repeatability con-ditions – Terms, methods, evaluation). Berlin: Beuth 2008. [11] Walther, S.: Untersuchungen zur Quantifizierung von
leicht-flüchtigen organischen Verbindungen (VVOC) in Luft mittels Gaschromatographie (Research on the quantification of vola -tile organic compounds (VVOC) in air by gas chromatography. Bachelor thesis. Beuth University of Applied Sciences Berlin, 2012.
[12] Richter, M.; Jann, O.; Horn, W.; Pyza, L.; Wilke, O.: System to generate stable longterm VOC gas mixtures of concentra -tions in the ppb range for test and calibration purposes. Gefahrstoffe – Reinhalt. Luft. Air Quality Control 73 (2013) No. 3, pp. 103-106.
